Received: 20 October 2016 | Revised: 20 January 2017 | Accepted: 28 January 2017 DOI: 10.1002/ece3.2860

ORIGINAL RESEARCH

Postglacial recolonization shaped the genetic diversity of the winter ( brumata) in Europe

Jeremy C. Andersen1 | Nathan P. Havill2 | Adalgisa Caccone3 | Joseph S. Elkinton1

1Department of Environmental Conservation, University of Massachusetts Abstract Amherst, Amherst, MA, USA Changes in climate conditions, particularly during the Quaternary climatic oscillations, 2 USDA Forest Service, Northern Research have long been recognized to be important for shaping patterns of species diversity. Station, Hamden, CT, USA For species residing in the western Palearctic, two commonly observed genetic pat- 3Department of Ecology & Evolutionary Biology, Yale University, New Haven, CT, USA terns resulting from these cycles are as follows: (1) that the numbers and distributions of genetic lineages correspond with the use of geographically distinct glacial refugia Correspondence Jeremy C. Andersen, Department of and (2) that southern populations are generally more diverse than northern popula- Environmental Conservation, University of tions (the “southern richness, northern purity” paradigm). To determine whether these Massachusetts Amherst, Amherst, MA, USA. Email: [email protected] patterns hold true for the widespread pest species the (Operophtera bru- mata), we genotyped 699 individual winter collected from 15 Eurasian coun- Present address Jeremy C. Andersen, Department of tries with 24 polymorphic microsatellite loci. We find strong evidence for the presence Environmental Science Policy and of two major genetic clusters that diverged ~18 to ~22 ka, with evidence that second- Management, University of California Berkeley, Berkeley, CA, USA ary contact (i.e., hybridization) resumed ~ 5 ka along a well-­established hybrid zone in Central Europe. This pattern supports the hypothesis that contemporary populations Funding information U.S. Forest Service, Grant/Award Number: 12- descend from populations that resided in distinct glacial refugia. However, unlike USDA-FS JV-11242303-096 and USDA-FS many previous studies of postglacial recolonization, we found no evidence for the 13-CA-11420004-236 “southern richness, northern purity” paradigm. We also find evidence for ongoing gene flow between populations in adjacent Eurasian countries, suggesting that long-­ distance dispersal plays an important part in shaping winter moth genetic diversity. In addition, we find that this gene flow is predominantly in a west-­to-­east direction, sug- gesting that recently debated reports of cyclical outbreaks of winter moth spreading from east to west across Europe are not the result of dispersal.

KEYWORDS biogeography, gene flow, glacial refugia, hybrid zone, invasive pests, outbreak waves

1 | INTRODUCTION oscillations (see references in Rousselet et al., 2010), and continue to be shaped by anthropogenic climate change (Svenning, Eiserhardt, Changes in climate conditions have long been recognized to be a major Normand, Ordonez, & Sandel, 2015). In the western Palearctic, nu- driver of range expansion, and local adaptation (Elias, Faria, Gompert, merous studies have shown that diversification has been promoted & Hendry, 2012). This is particularly true for many species dwelling during periods when species’ distributions were restricted to geo- in temperate regions whose recent evolutionary histories have been graphically isolated glacial refugia in the southern peninsulas and/ shaped by changes in their distribution during the Quaternary climatic or alpine regions of the European continent (see reviews by Hewitt,

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2017 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.

.  Ecology and Evolution. 2017;7:3312–3323 www.ecolevol.org | 3312 ANDERSEN et al | 3313

1996, 2000; and Schmitt, 2007). For contemporary populations, the Europe (Wint, 1983), and has long been studied as a model organism effects of these periods of isolation are often observed by the pres- for studies of local adaptation (Tikkanen & Lyytikainen-­Saarenmaa, ence of distinct genetic lineages that correspond with the numbers 2002; Tikkanen, Woodcock, Watt, & Lock, 2006; Van Dongen, and locations of the glacial refugia utilized by a particular species Matthysen, & Dhondt, 1996), and population ecology (Hassell, 1968; (Taberlet, Fumagalli, Wust-­Saucy, & Cosson, 1998). Macphee, Newton, & McRae, 1988; Varley & Gradwell, 1960, 1968), In addition to promoting patterns of diversification, the Quaternary and has been at the center of an ongoing debate in regards to whether climatic oscillations have also had profound impacts on genetic varia- cyclical outbreaks of geometrid moths (including winter moth) move tion within populations. One common finding for postglacial recolo- across western Eurasian from east to west approximately every nization studies is that populations in geographic regions that acted 10 years (Tenow et al., 2013; but see Jepsen, Vindstad, Barraquand, as glacial refugia during the glacial maximums are more genetically Ims, & Yoccoz, 2016 and Tenow, 2016). In addition, this species has diverse than populations in regions that have been recolonized (i.e., also been the subject of much genetic research, including population the “southern richness, northern purity” paradigm; Hewitt, 2000; structure (Leggett et al., 2011; Van Dongen, Backeljau, Matthysen, & Canestrelli, Cimmaruta, Costantini, & Nascetti, 2006; Babik et al., Dhondt, 1998), hybridization (Elkinton, Liebhold, Boettner, & Sremac, 2009; Rousselet et al., 2010; Wielstra et al., 2013; Tison et al., 2014; 2014; Elkinton et al., 2010; Havill et al., 2017), and a draft genome for Mezzasalma et al., 2015; Vitales, Garcia-­Fernandez, Garnatje, Pellicer, this species was recently published (Derks et al., 2015). Yet, the only & Valles, 2016). As populations of species recolonized temperate re- continent-­scale study of winter moth phylogeography (Gwiazdowski, gions, they likely extended their distributions in a stepping-­stone Elkinton, Dewaard, & Sremac, 2013) found little evidence for geo- manner whereby multiple genetic bottleneck events were associated graphically distinct genetic lineages when the mitochondrial locus cy- with continued northward range expansions (e.g., Tison et al., 2014). tochrome oxidase I (COI) was analyzed—although it did find support However, during these cycles genetic lineages that were reproduc- for a division between northern (i.e., Norway, Scotland, and Sweden) tively isolated during glacial maximums might also have subsequently and southern European (i.e., all other sampled locations) populations. come into secondary contact with one another during periods of recol- This result was unexpected, given ample evidence of local adapta- onization and the signatures of these secondary contact events have tion by winter moth populations (Tikkanen & Lyytikainen-­Saarenmaa, been shown by the presence of detectible hybrid zones (Hewitt, 2000; 2002; Tikkanen et al., 2006; Van Dongen et al., 1996), and the fact Schmitt, 2007; Schmitt & Müller, 2007). Hybridization—here used in that females are flightless and that males are considered poor dispers- the context of admixture between distinct populations (Harrison & ers (Van Dongen et al., 1996). Larson, 2014)—has long been known to play an important role in shap- Therefore, we were interested if a recently developed set of 24 ing patterns of host-­associations (e.g., Feder et al., 2003), and has the highly variable microsatellite loci (Havill et al., 2017) could be used to potential to greatly increase measures of genetic diversity (Verhoeven, further elucidate patterns of genetic diversity for populations of win- Macel, Wolfe, & Biere, 2011) and even reverse the process of specia- ter moth. Specifically we were interested the following: (1) whether tion (Seehausen, Takimoto, Roy, & Jokela, 2007). How hybridization individual winter moths could be assigned to distinct genetic clusters might affect the “southern richness, northern purity” paradigm, how- using Bayesian clustering analyses and genetic distance methods and ever, is unclear. In part this may be because the strength of the sig- whether these genetic groupings corresponded with geographically nal of this paradigm may vary depending on the recolonization paths distinct glacial refugia, (2) whether divergence times between con- taken by a species, and the geographic locations of secondary contact temporary populations in regions that likely acted as glacial refugia zones (see Hewitt, 2000 and Schmitt, 2007). corresponded with isolation during the last glacial maximum (LGM), Species of have been crucial for examining both of (3) whether winter moth populations display evidence of decreasing the above biogeographic patterns. This is could be due to the fact genetic diversity with increases in latitude, and (4) whether contem- that many species of Lepidoptera have the ability to disperse long dis- porary gene flow could be detected and whether the direction of gene tances and can quickly fill open niche space at the species-­level, while flow can inform the debate about cyclical outbreaks of geometrid dispersal distances for individuals are generally quite small, allowing moths. for the preservation of biogeographic structure (Schmitt, 2007). These short dispersal distances may also be important for the promotion of 2 | MATERIALS AND METHODS local adaptation (Tison et al., 2014), and recent studies of Lepidoptera have also highlighted the importance of micro-refugia­ (von Reumont, 2.1 | Study species and sampling strategy Struwe, Schwarzer, & Misof, 2012), postcolonization hybridization (Schmitt & Müller, 2007), and colonization from refugia outside of Winter moth is broadly distributed across western Asia, Europe Europe (Habel, Lens, Rodder, & Schmitt, 2011) for patterns of Eurasian (Tenow et al., 2013; Troubridge & Fitzpatrick, 1993), and North biogeography. However, one species of Lepidoptera for which eluci- Africa (Ferguson, 1978; Mannai, Ezzine, Nouira, & Ben Jamaa, 2015), dating historical biogeographic patterns has been particularly elusive and is a non-­native invasive pest in North America (reviewed in is the winter moth, Operophtera brumata (Lepidoptera: Geometridae). Elkinton et al., 2014; Elkinton, Boettener, Liebhold, & Gwiazdowski, This species has a broad range of woody host plants including, but 2015). Larvae of this species hatch in early spring, often just prior not limited to, oak (Quercus), maple (Acer), and birch (Betula) trees in to or synchronized with budburst (Buse & Good, 1996; Tikkanen & ANDERSEN et al | 3314

Lyytikainen-­Saarenmaa, 2002; Tikkanen et al., 2006; Van Dongen, Therefore, we estimated the probability of assignment (Z) that an indi- Backeljau, Matthysen, & Dhondt, 1997). Once emerged, larvae cause vidual could be classified as belonging to one of these two clusters or considerable damage to their hosts, which include a broad range of to hybrid categories (F1, F2, or backcrosses), using the software pro- deciduous (Wint, 1983) and evergreen (Watt & Mcfarlane, 1991) gram NewHybrids v.1.1.b3 (Anderson, 2008; Anderson & Thompson, trees and shrubs. After pupation in the leaf-­litter, flightless adult fe- 2002). Four independent runs, each of 1 million generations, discard- males and winged adult males emerge during November or December ing the first 100,000 burn-­in generations, were analyzed using random (in most regions) and females attract mates using a sex pheromone starting values, and uniform priors for the estimates of θ (i.e., allele fre- (Roelofs et al., 1982) before depositing between 150–350 eggs that quencies) and π (i.e., mixing proportions). Results were then averaged overwinter until the following spring (Elkinton et al., 2015). across runs, and visualized using ArcMap v.10.3.1 (Esri, Redlands, CA, For this study, male moths were collected at 44 locations in Eurasia USA). (Table S1, Figure S1) using sex pheromone-­baited traps deployed in tree canopies that attract winter moth and several closely related con- 2.3.2 | Population genetic statistics and geners (Elkinton, Lance, Boettner, Khrimian, & Leva, 2011; Elkinton genetic distances et al., 2010). Moths were collected from the traps and then placed in glassine envelopes (Uline Corporation, USA), before storage at −80°C. For all localities from which ≥10 individuals were successfully geno- typed, we calculated standard population genetic summary statistics, including: the number of individuals genotyped (n), the average num- 2.2 | DNA extraction and amplification ber of alleles per locus (Na), the effective number of alleles (Eff_Na),

Whole genomic DNA was extracted from individual male winter moths the average observed heterozygosity across loci (Ho), the average in- using the Qiagen DNeasy Kit (Qiagen Corporation, Netherlands) fol- breeding coefficient (GIS), and the presence of deviation from Hardy– lowing the manufacturer’s instructions. Prior to DNA extraction, the Weinberg Equilibrium (HWA) using GenoDive v.2.0b27 (Meirmans & tip of the abdomen (genitalia are used for morphological species Van Tienderen, 2004). In addition, we tested each locus-­pair within identification) and wings were removed and stored as vouchers for populations for evidence of Linkage-­Disequilibrium (LD) using most of the samples. Vouchers were deposited at the Yale Peabody GenePop (Raymond & Rousset, 1995; Rousset, 2008), and we esti- Museum of Natural History, New Haven Connecticut, USA. Twenty-­ mated the frequency of null alleles present for each locus and popula- four polymorphic microsatellites were then amplified from each indi- tion using FreeNA (Chapuis & Estoup, 2007). The degree of genetic vidual using primers and protocols presented in Havill et al. (2017). differentiation (FST) among populations was calculated (accounting for Genotyping was conducted at the DNA Analysis Facility on Science the presence of null alleles) in FreeNA, and Jost’s estimator of actual

Hill at Yale University using a 3730xl DNA Analyzer (Thermo Fisher differentiation (Dest) was calculated using Smogd v.1.2.5 (Crawford, Scientific, USA), and fragment lengths were scored in comparison with 2010). To examine whether genetic differentiation was correlated the GeneScan 500 LIZ size standard (Thermo Fisher Scientific) using with geographic distances, the presence of isolation-­by-­distance was the microsatellite plug-­in in the software program Geneious v.8.1.6 estimated by performing linear-­regression of FST / (1-­FST), calculated (Kearse et al., 2012). in GenePop, versus the linear distance between populations (km) in R v. 3.1.3 (R Core Team 2015). Finally, to examine historical relation- ships among populations a “NeighborNet” network was reconstructed 2.3 | Winter moth postglacial recolonization with the above null-­allele corrected FST estimates with the program After genotyping, individuals from which ≥20 microsatellite loci SplietRetree v.4.14.2 (Huson & Bryant, 2006). were successfully amplified were included in subsequent analyses. Genotype data for each included individual is provided as a single-­line 2.3.3 | Historical demography formatted file titled (Appendix S2). To determine whether divergence timing for contemporary Eurasian populations of winter moth correspond with genetic isolation dur- 2.3.1 | Bayesian genetic clustering ing the LGM, population history parameters were estimated using Previous postglacial recolonization studies have found that the Approximate Bayesian Computation (ABC) with the software program

Quaternary climatic cycles have played an important role in shaping DiyABC v.2.1.0 (Cornuet et al., 2008, 2014). For these analyses we genetic diversity. This diversity is evident by the presence of two to included samples from Spain, Serbia, and Georgia as representative of three distinct genetic lineages in Eurasia (Hewitt, 2000; and Schmitt, three potential glacial refugia (Iberaian, Balkan, and Caucasus, respec- 2007), with well-­established hybrid zones where these lineages have tively). Due to limited sampling from Italy (Italian refugium), we did come into secondary contact with each other (see Hewitt, 2000; not include samples from this region in our analysis. In addition, we

Figure 3). Preliminary analyses of our dataset using SetEuceture v.2.3.2 included samples from Germany to represent the product of postgla- (Falush, Stephens, & Pritchard, 2003; Pritchard, Stephens, & Donnelly, cial recolonization. We evaluated eight different recolonization sce- 2000) indicated that individual winter moths could be assigned to narios (Figure S2). Three scenarios included simple bifurcating trees an optimal number of two distinct genetic clusters (Appendix S1). that differed in whether the German population was derived from ANDERSEN et al | 3315 the Spanish, Serbian, or Georgian populations, and five scenarios that are reported as significant. Average rates of gene flow between east- included the German population being derived from admixture. All ern (Austria, Czech Republic, Georgia, Poland, Serbia, Slovakia) and scenarios included multiple population size parameters to allow for western (England, France, Germany, Italy, Norway, Scotland, Spain, changes in population sizes following splitting and/or merging events. Sweden, Switzerland) European countries were then calculated. The Prior information for all parameters, including minimum and maximum difference between these rates was used as the basis for simulations values, as well as distributions, is presented in Table S2. Default muta- to determine if one or the other populations would eventually become tion model parameters were used, except for the following: we set ubiquitous and if so in how many generations. Simulations were car- −5 the minimum mean mutation rate to 1 × 10 , and maximum values ried out using the software program PopG v. 4.03 (© University of for the Mean and Individual locus coefficient P’s to 1.0. Visualization Washington and Joseph Felsenstein), under a simple two allele model of preliminary results based on principal component analyses (PCAs) with equal starting frequencies and the observed difference in migra- within DiyABC indicated that these changes in the mutation rate and tion rates. Ten independent simulations were conducted; each with the Mean and Individual locus coefficient P’s improved the shape of random starting seeds, and the number of generations to fixation of the cloud of simulated datasets. In addition, visualization of prelimi- one of the alleles averaged across runs was recorded. nary results based on PCAs indicated that the removal of four loci (02339, 00925, 02191, and 12042) also improved the shape of the 3 | RESULTS cloud of simulated datasets. These four loci each had allelic ranges ≥120 repeat units (i.e., the number of tandem repeats). Given that 3.1 | Winter moth postglacial recolonization lepidopteran genomes have high rates of duplication events (Edger et al., 2015; You et al., 2013), and that the development of microsatel- 3.1.1 | Bayesian genetic clustering lite loci for these taxa have been difficult due to high frequencies of null alleles and associations with transposable elements (see citations After filtering the dataset to include only those individuals from which in Sinama et al., 2011), it is possible that there might be an association ≥20 microsatellite loci were genotyped, analyses included 669 individ- between these four loci and transposable elements and/or duplica- ual moths. The proportional assignment of individuals by NewHybrids tion events. We will continue to monitor the appropriateness of using is presented in Table S1 and summarized by country in Figure 1. These these four loci. We then generated a reference table by simulating results showed that 145 individuals could be assigned to Cluster 1 8 million datasets (1 million datasets per scenario). Pre-­evaluations (123 individuals with high probability of assignment Z ≥ 0.8, and 22 of the simulated datasets were conducted using PCA, and scenarios individuals with moderate probabilities of assignment 0.5 ≤ Z < 0.8). were compared using both the “Direct” (i.e., the proportion of simu- Four hundred and thirty-­four individuals were assigned to Cluster 2 lated datasets from each scenario “closest” to the sample dataset) and Cluster 2 (386 with high probability of assignment and 47 with mod- “Logistic Regression” (i.e., a logistic regression analysis of the prob- erate probability of assignment). In addition, 78 individuals were ability of deviations between summary statistics calculated for the identified as hybrids (13 with high probability of assignment and simulated datasets and the sample dataset) tests in DiyABC. 65 with moderate probabilities of assignment). All hybrids were as- signed to the F2 hybrid class. Thirteen individuals were classified as “unassigned” due to low probabilities of assignment (Z < 0.5 to any 2.3.4 | Genetic diversity and latitude category). In general, assignments showed a clear distinction between

To examine whether measures of genetic diversity (Na, Eff_Na, Ho, and eastern and western Europe, with moths from Georgia, Poland, and

GIS) were correlated with latitude, we constructed generalized addi- Serbia being primarily assigned to Cluster 1 (100%, 97%, and 88%, tive models (GAM) using the R package “mgcv” (Wood, 2011). Results respectively), and individuals from Spain, Scotland, and England being were then visualized using the “plot.gam” function. primarily assigned to Cluster 2 (100%, 100%, and 96%, respectively). In addition, while the overall number of hybrids was low, the greatest proportions of hybrid individuals were found in Austria, Sweden, and 2.4 | Contemporary rates of gene flow the Czech Republic (39%, 33%, and 32%, respectively), roughly cor- The rates of contemporary gene flow among populations in Eurasia responding with a previously identified hybrid zone in central Europe were examined by estimating the proportion of migrant individuals in (see Figure 3 in Hewitt, 2000). each population using the software program BayesAss v.4.0 (Wilson & Rannala, 2003). To reduce the number of pair-­wise comparisons, indi- 3.1.2 | Population genetic statistics and viduals were grouped by country. Gene flow rates were then estimated genetic distances for all 205 possible comparisons with four independent analyses, each with a random starting seed, runtimes of 10 million generations, Population genetic statistics for 26 locations from which ≥10 indi- burn-­in periods of 1 million generations, and by setting all mixing pa- vidual moths were sampled showed that the average number of al- rameters to 0.8. Results were then summarized across analyses using leles per locus was Na = 8.63, the average effective number of alleles etracer v.1.6.0 (Rambaut & Drummond, 2007), and all migration rates per locus was Eff_Na = 4.34 the average observed heterozygosity whose 95% confidence intervals (mean ± [1.96 × SE]) did not include 0 was Ho = 0.58, and the average inbreeding coefficient was GIS = 0.17 ANDERSEN et al | 3316

20°0'0"W 10°0'0"W 0°0'0" 10°0'0"E 20°0'0"E 30°0'0"E 40°0'0"E 50°0'0"E 60°0'0"E Legend Norway Significant geneflow: 0.005 to 0.009 0.010 to 0.049 0.050 to 0.099 60°0'0"N 0.100 to 0.200

Genetic Clusters:

Sweden Z ≥ 0.5 to Western Cluster Scotland 50°0'0"N Z ≥ 0.5 to Eastern Cluster

Z ≥ 0.5 to F2 Hybrids

Z < 0.5 to any Cluster

Poland Glacial Refugia: 50°0'0"N England Germany C.R. as per Schmitt (2007) Fig. 3

Slovakia

Georgia Switzerland Austria 40°0'0"N France

Serbia Italy Spain 40°0'0"N

07370K40 1,480 ilometrse

0°0'0" 10°0'0"E 20°0'0"E 30°0'0"E 40°0'0"E

FIGURE 1 Proportional assignment of individuals to hybrid categories, and contemporary gene flow rates among Eurasian countries. Pie charts, representing the proportion of individuals in each country assigned to different hybrid categories calculated with NewHybrids (Z ≥ 0.5 to Cluster 1 = Black; Z ≥ 0.5 to Cluster 2 = White; Z ≥ 0.5 to F2 Hybrid = Dark Gray; Z < 0.5 to any category = Light Gray), were generated in ArcMap v.10.3.1 and visualized using the Europe Albers Equal Area Conic projection. The location of each chart is placed approximately in the center of each country. The label for the Czech Republic has been abbreviated “C.R.” Gene flow rates are presented as the proportion of individuals that are likely migrants from a different Eurasian country as calculated by BayesAss. The thickness of the curved lines represents the proportion of migrants (see legend, top right), and the arrowhead indicates the direction of gene flow. The distribution of glacial refugia during the LGM (diagonal shading) is drawn as per Figure 3 in Schmitt (2007)

(Table 1). The average presence of null alleles across loci was 6.7%. 3.1.3 | Historical demography All locations showed significant (p ≤ .05) deviations from HWE, while only 13 of the 276 pair-­wise comparisons showed significant (p ≤ .05) Comparisons of the posterior probabilities of the scenarios included in evidence of LD between loci after using the Benjamini–Hochberg the DiyABC analysis indicated that Scenarios 4 and 7 were the best fit false-­discovery method (Benjamini & Hochberg, 1995). for the data (Figure S3). Under the Direct approach simulations based Among populations, the average values for pair-­wise differentia- on Scenario 7 were consistently the most supported, whereas under tion for FST (corrected for the presence of null alleles) was 0.060 and for the Logistic Regression approach support moves from Scenario 4 to

Dest was 0.091 (Table S3). Genetic differentiation also showed strong Scenario 7 with increasing distance from the sample data (Figure S3). patterns of isolation-­by-­distance (p = 2.2 × 10−16, r2 = .5235, df = 323) These two scenarios differ in regards to whether the Serbian popula- among sampled populations. Network analysis indicated that most tion is sister to the Georgian or the Spanish populations, yet both sce- populations could be divided into two large subnetworks (Figure 2). narios indicate that the German populations are derived from recent One of these subnetworks included populations from Austria, Czech admixture between the populations that recolonized Europe from Republic, Poland, Serbia, and Slovakia (labeled “Eastern Europe”), while the Iberian and Balkan refugia. The branching pattern observed in the other subnetwork included populations from England, France, Scenario 7 also matches the results observed from the Bayesian clus- Germany, Italy, and Switzerland (labeled “Western Europe”). In ad- tering and distance analyses (i.e., Georgia being most closely related to dition, populations of moths from Scotland, Spain, Norway, and the eastern European populations). Results for all estimated parameters Republic of Georgia were each placed at the ends of long branches. for these two scenarios are presented in Table 2, and the mean values ANDERSEN et al | 3317

TABLE 1 Population genetic summary Population n N Eff_N H G HWE %Hybrids statistics calculated using GenoDive, a a o IS including: the number of individuals Austria, Neulengbach 29 10.125 4.841 0.610 0.167 0.001 37.93 genotyped (n), the average number of Austria, Vienna 35 10.625 4.802 0.623 0.113 0.001 40.00 alleles per locus (N ), the effective number a Czech Republic, Prague 19 8.083 4.505 0.535 0.251 0.001 31.58 of alleles (Eff_Na), the average observed England, Farnham 48 10.583 4.149 0.566 0.185 0.001 2.08 heterozygosity (Ho), the inbreeding coefficient (GIS), the probability (p) of France, Bédarieux 16 7.042 3.966 0.579 0.183 0.001 0.00 deviation from Hardy–Weinberg France, Haguenau 31 9.292 4.451 0.619 0.145 0.001 12.90 Equilibrium (HWE), and the percent of individuals in the population classified as France, Marcillac 29 10.500 4.876 0.645 0.128 0.001 0.00 hybrids using NewHybrids (%Hybrids) France, Onet L’eglise 20 6.875 3.732 0.481 0.270 0.001 0.00 France, Rennes 23 8.458 4.334 0.587 0.156 0.001 0.00 Georgia, Tbilisi 27 7.500 3.519 0.519 0.117 0.001 0.00 Germany, Bühren 30 10.500 4.695 0.614 0.170 0.001 13.33 Germany, Göttingen 30 10.000 4.803 0.633 0.143 0.001 16.67 Germany, 30 10.375 4.694 0.668 0.111 0.001 13.33 Reinhardshagen Germany, Schlüchtern 18 8.542 4.494 0.597 0.181 0.001 11.11 Italy, Tregnago 15 7.083 4.041 0.531 0.242 0.001 15.38 Norway, Tromsø 29 9.125 4.469 0.601 0.173 0.001 6.67 Poland, Bialostocka 20 9.292 5.106 0.608 0.152 0.001 0.00 Scotland, Banchory 10 5.042 2.89 0.539 0.089 0.011 0.00 Scotland, Torphins 30 7.917 3.459 0.526 0.160 0.001 0.00 Serbia, Belgrade 15 7.208 4.471 0.558 0.197 0.001 6.67 Serbia, Pančevo 37 11.792 6.11 0.570 0.218 0.001 8.11 Slovakia, Banská 20 8.542 4.665 0.537 0.248 0.001 25.00 Štiavnica Spain, Lugo 16 4.708 2.583 0.448 0.138 0.001 0.00 Sweden, Uppsala 21 8.792 4.69 0.645 0.145 0.001 33.33 Switzerland, Delémont 21 8.250 4.176 0.651 0.086 0.001 4.76 Switzerland, Malettes 19 8.208 4.179 0.625 0.133 0.001 10.53

for divergence times and population sizes are presented graphically in 3.1.4 | Genetic diversity and latitude Figure 3. In both scenarios, the German and Serbian effective popula- tion sizes are significantly larger (non-­overlapping 95% CIs) than the Results from the GAM analyses found no significant relationships

Georgian and Spanish populations. The mean estimates for population between measures of genetic diversity (Na, GIS, and Ho) and latitude. sizes were all larger than the estimates for the ancestral population However, Eff_Na and latitude showed a marginally significant rela- 2 (NA5), although again these were only significant for the German tionship (p = .0838, r2 = .361, deviance explained = 55.3%; Figure and Serbian populations, and there were no significant differences be- S4). tween the ancestral population 1 (NA3) and any contemporary popu- lations suggesting that population sizes have remained somewhat 3.2 | Contemporary rates of gene flow stable through time. In both scenarios 4 and 7, the mean estimate for when admixture between the Spanish and Serbian populations oc- Included in the gene flow analyses were all individuals from Austria curred was ~5 ka (Scenario 4 = 5.19 ka [95% CI = 1.32 ka, 13.8 ka]; (n = 64), Czech Republic (n = 19), England (n = 48), France (n = 124), Scenario 7 = 4.9 ka [95% CI = 1.25 ka, 12.3 ka]), and the mean es- Georgia (n = 27), Germany (n = 108), Italy (n = 15), Norway (n = 30), timates for the divergence time between the eastern and western Poland (n = 29), Scotland (n = 40), Serbia (n = 52), Slovakia (n = 20), populations was ~ 20 ka (Scenario 4 t2 = 18.1 ka [95% CI = 6.82 ka, Spain (n = 20), Sweden (n = 21), and Switzerland (n = 52). Visual in-

29.1 ka]; Scenario 7 t3 = 21.3 ka [95% CI = 8.63 ka, 44.5 ka]). In both spection of the four BayesAss runs using etracer indicated that after scenarios, the German population is derived from near equal propor- 10 million generations all runs had converged on similar estimates tions of admixture between the Spanish and Serbian populations for migration rates. After summarizing across runs, 15 of 205 pos- (ra = 0.462 [95% CI = 0.047, 0.94] and ra = 0.412 [95% CI = 0.065, sible migration routes showed evidence of significant gene flow (i.e., 0.866] for scenarios 4 and 7, respectively). 95% CIs did not include 0) between populations (Table S4; Figure 1). ANDERSEN et al | 3318

Western Europe 0.001 Eastern Europe

Scotland Tregnago Onet L’eglise Marcillac G-S-D-MBührenReinhardshagen

Torphins Uppsala Prague Neulengbach tiavnica m Banchory Farnha Rennes Banská Vienna Panč Belgrade Haguenau Bédarieux evo Bialostocka

Spain Norway Lugo Tromsø Republic of Georgia

0.001

Tbilisi

Tromsø

Lugo

Tbilisi

FIGURE 2 Network Analysis based on FST (corrected for the presence of null alleles). Branches are drawn proportional to the scale bar in the upper left corner. Branches leading to Tbilisi, Georgia and Lugo, Spain have been “broken” in the main figure, but for reference these branches are shown intact in the inset figure. Four sampled locations, Göttingen, Schlüchtern, Delémont, and Malettes had distances of FST = 0.00, and the node with these four populations is indicated with the label “G-­S-­D-­M”

Germany was the country with the highest number of significant con- 4 | DISCUSSION nections, being the source of migrants to six other countries, as well as the recipient of migrants from France. Austria and France each 4.1 | Winter moth postglacial recolonization had five connections (each with four sources and one recipient in- teraction), while Spain, Scotland, and Georgia showed no evidence We find that populations of winter moth can be broadly assigned to two of contemporary gene flow. Only Germany and France showed evi- geographically and genetically distinct populations (Figure 1), which di- dence of reciprocal gene flow, although gene flow was greater from verged ~ 20 ka (Figure 3). While this finding differs slightly from previous Germany to France than the reverse. Of the 15 connections, 13 of evidence that suggested northern winter moth populations might be dis- them were directed in an easterly direction (e.g., east, north-­east, tinct from southern populations (Gwiazdowski et al., 2013), the observed south-­east, etc.), while only two were directed in a westerly direc- results are consistent with many previous studies that have highlighted tion (Figure 1). The average dispersal rate from western to eastern the importance of the Quaternary climate cycles in shaping patterns of countries was m = 0.0125, while for eastern to western countries it genetic diversity and geographic distributions for other lepidopteran was m = 0.0070, with a difference in the average rate of migration of species in Europe and western Asia (e.g., Hewitt, 2000; Schmitt, 2007; m = 0.0055. Simulations run in PopG using the observed difference in Schmitt, Rober, & Seitz, 2005; Schmitt & Seitz, 2001a,b; Wu et al., 2015). migration rates found the average time to fixation was 7,000 genera- For winter moth, our findings suggest that populations became re- tions (±3,830 generations). productively isolated in glacial refugia located both in the Iberian and ANDERSEN et al | 3319

TABLE 2 Parameter estimates (mean and 95% Confidence Intervals [CI]) from DiyABC for Scenarios 4 and 7 based on 1 million simulated datasets each. These estimates include; current effective population sizes in Georgia (N1), Serbia (N2), Spain (N3), and Germany (N4), as well as effective population sizes for ancestral population 1 (NA3) and ancestral population 2 (NA5). In addition, estimates are shown for the proportion of admixture (ra), time of divergence (t1, t2, and t3); as well as the average mutation rates (μ-­mic), the average coefficient of P (p-­mic), and the average SNI rates (sni-­mic) for the microsatellite loci

Scenario 4 Scenario 7

Mean 95% CI Lower 95% CI Upper Mean 95% CI Lower 95% CI Upper

Georgia 1.53E+04 7.24E+03 2.77E+04 1.85E+04 9.46E+03 2.97E+04 Serbia 5.62E+04 3.11E+04 8.60E+04 6.20E+04 3.68E+04 8.91E+04 Spain 1.14E+04 5.31E+03 2.11E+04 1.20E+04 5.32E+03 2.30E+04 Germany 6.75E+04 3.98E+04 9.32E+04 6.52E+04 3.96E+04 9.17E+04 Ancestral population 1 3.14E+04 2.27E+03 7.57E+04 2.02E+04 9.40E+02 7.05E+04 Ancestral population 2 2.75E+03 7.74E+01 8.86E+03 3.35E+03 1.08E+02 9.36E+03 ra 4.62E-­01 4.73E-­02 9.40E-­01 4.12E-­01 6.52E-­02 8.66E-­01 t1 5.19E+03 1.32E+03 1.38E+04 4.90E+03 1.25E+03 1.23E+04 t2 1.81E+04 6.82E+03 2.91E+04 2.07E+04 9.99E+03 2.93E+04 t3 2.08E+04 8.71E+03 4.21E+04 2.13E+04 8.63E+03 4.45E+04

μ-­mic 1.19E-­04 3.38E-­05 2.52E-­04 1.25E-­04 3.46E-­05 2.96E-­04 p-­mic 6.44E-­01 2.34E-­01 9.75E-­01 6.28E-­01 2.26E-­01 9.71E-­01 sni-­mic 1.06E-­07 1.00E-­08 7.70E-­07 1.15E-­07 1.00E-­08 8.49E-­07

Balkan peninsulas and in the Caucasus region, as a result of the ex- Arroyo, Jordano, & Petit, 2003; Jaarola & Searle, 2002), however the pansion of the glaciers to their glacial maximums ~ 26 ka (Clark et al., area connecting the Caucasus region to continental Europe has gone 2009). During this period, populations in the Balkan Peninsula and through significant changes in forest cover since the last glacial max- the Caucasus evolved independently from populations in the Iberian imum (Adams, 1997), and it is possible that these changes eventually Peninsula as evident by the level of genetic divergence observed be- prevented gene flow between winter moth populations in eastern tween contemporary populations (Average FST = 0.060). This level of Europe and the Caucasus region. divergence is similar to that which has been observed for other species Genetic distance analyses also revealed that populations from two of Lepidoptera (e.g., Schmitt & Seitz, 2001a; FST = 0.060), but less than countries, Norway and Scotland, that were recolonized after the re- that which has been observed for species with recognized subspecies treat of the glaciers showed additional evidence of genetic differen-

(e.g., Schmitt & Seitz, 2001b; FST = 0.149). Beginning ~20 ka the glaciers tiation (Figure 2). In northern Norway, winter moth recently became began to retreat (Clark et al., 2009), and during the subsequent refor- established (~ 125 ya), and its introduction has caused extensive de- estation of the European continent, these two lineages expanded north- foliations due to frequent outbreaks (Vinstad et al., 2013). Our gene ward, coming into secondary contact ~ 5 ka along a well-­documented flow analyses (Figure 1) indicated the presence of significant gene flow hybrid zone in Central Europe (see Figure 3 in Hewitt, 2000). from populations in France to this region, suggesting that France might Following the northward recolonization of the European con- be the source of this invasive population. If this is correct, then the tinent, populations in Spain and Georgia—regions that acted as gla- long branch linking the Norwegian population to the other western cial refugia during the last glacial cycle—likely became reproductively European populations (Figure 2) may be the result of a genetic bottle- isolated in these regions as evident by the long branches connecting neck created during the invasion process. However, further sampling these populations to other European populations (Figure 2). For the of the Fennoscandia region is required to comment further on the ori- Spanish population, this result might highlight the importance of phys- gins of this population. Similarly, the long branch connecting the pop- ical barriers, such as the Pyrenees, in promoting reproductive isolation, ulations of winter moth in Scotland to the other Western European as seen for the pine processionary moth, Thaumetopoea pityocampa populations is also interesting given that populations in Scotland have (Rousselet et al., 2010). Similarly, the distance of the branch leading become a serious pest of Sitka spruce (Hunter, Watt, & Docherty, 1991; to the Georgian population might also highlight the importance of Stoakley, 1985), suggesting that this long branch might be the result of geographic isolation, a pattern similar to that found for white oaks micro-­evolutionary processes as a result of this significant host shift. (Dumolin-­Lapègue, Demesure, Fineschi, LeCorre, & Petit, 1997), one of the primary hosts for winter moth. The Caucasus region has long 4.1.1 | Genetic diversity and latitude been recognized as having played an important role as a refugium during glacial cycles (e.g., Babik et al., 2005; Beck, Schmuths, & Schaal, One common finding for postglacial recolonization studies is that 2007; Dubey, Zaitsev, Cosson, Abdukadier, & Vogel, 2006; Hampe, populations in geographic regions that acted as glacial refugia ANDERSEN et al | 3320

Scenario 4 Scenario 7 t3 ≈ 21.3 ka (8.63 ka, 44.5 ka) t3 ≈ 20.8 ka t2 ≈ 20.7 ka Ancestor 2 (8.71 ka, 42.1 ka) Ancestor 2 (9.99 ka, 29.3ka) Ancestor 1 Ancestor 1

Georgia t2 ≈ 18.1 ka Georgia Serbia (6.92 ka, 29.1 ka) Serbia

Spain Spain Germany Germany

t1 ≈ 5.2 ka t1 ≈ 4.9 ka ra 1-ra (1.32 ka, 13.8 ka) ra 1-ra (1.25 ka, 12.3 ka)

0 0 Spain GermanySerbia Georgia Spain GermanySerbia Georgia

FIGURE 3 The top two scenarios for diversification and recolonization identified by simulations in DiyABC. Populations in each scenario were chosen to represent different glacial refugia (e.g., Spain = Iberian Refugium, Serbia = Balkan Refugium, Georgia = Caucasus Refugium), as well as a population that would have established after the northward retreat of the glaciers following the LGM (e.g., Germany). Both of these supported scenarios include an admixture event between the Serbian and Spanish populations giving rise to the German population. Mean estimates for merging (t1) and divergence (t2 and t3) events are shown along the y-­axis along with 95% confidence intervals. The thicknesses of branches are drawn proportional to the mean estimated population size. Estimates for all parameters, including 95% confidence intervals, are presented in Table 2. All eight tested scenarios is show in Figure S2 during the LGM are more genetically diverse than populations in 4.2 | Contemporary rates of gene flow recolonized regions (Babik et al., 2009; Canestrelli et al., 2006; Hewitt, 2000; Mezzasalma et al., 2015; Rousselet et al., 2010; Another interesting finding from our study involves the level and Tison et al., 2014; Vitales et al., 2016; Wielstra et al., 2013). Here direction of gene flow among contemporary European populations. we find some evidence that measures of genetic diversity might Recent studies have explored whether outbreaks of winter moth be greater in middle European latitudes (Figure S4), a region that occur on a cyclical basis, suggesting a cycle length of approximately is experiencing secondary contact (i.e., hybridization; Figure 1). 10 years during which winter moths move across Europe predomi- Given that hybridization has the potential to provide populations nantly from east to west (Tenow et al., 2013; but see Jepsen et al., with additional genetic diversity, and that rates of hybridization are 2016 and Tenow, 2016). These studies have been important contri- much higher than previously suspected (Allendorf, Leary, Spruell, butions to our understanding of the incidence and causes of cyclic & Wenburg, 2001; Harrison & Larson, 2014) particularly among population dynamics and of damaging outbreaks, however the insect taxa (Schwenk, Brede, & Streit, 2008), secondary contact factors responsible for these outbreaks remain unclear. Dispersal following postglacial recolonization may play an important, and un- may be an important driver of outbreaks of herbivorous derreported, role in promoting the genetic diversity of populations (Liebhold, Koenig, & Bjornstad, 2004; Peltonen, Liebhold, Bjornstad, in western Eurasia. & Williams, 2002), although it is but one of numerous biotic and ANDERSEN et al | 3321 abiotic factors that can be responsible for the observed patterns CONFLICT OF INTERESTS (Hunter & Elkinton, 2000; Price & Hunter, 2015; Riolo, Rohani, & The authors declare no conflict of interests. Hunter, 2015). Previous studies have highlighted the ability of winter moth populations to undergo long-­distance dispersal (Leggett et al., 2011), likely through a process known as ballooning where larvae REFERENCES secrete a silky thread and are subsequently wind-­dispersed (Elkinton Adams, J. M. (1997). Global land environments since the last interglacial. et al., 2015). Therefore, long-­distance dispersal by larvae may pro- Oak Ridge National Laboratory, TN, USA. Retrieved from: http://www. vide a mechanism that drives these outbreak waves. If this were the esd.ornl.gov/ern/qen/nerc.html Allendorf, F. W., Leary, R. F., Spruell, P., & Wenburg, J. K. (2001). The prob- case, then we might expect levels of gene flow to be primarily in lems with hybrids: Setting conservation guidelines. Trends in Ecology the east to west direction (the proposed direction of the outbreak and Evolution, 16, 613–622. waves), however; our results suggest that long-­distance dispersal Anderson, E. C. (2008). Bayesian inference of species hybrids using multi- occurs primarily in the opposite direction (from west to east). In locus dominant genetic markers. Philosophical Transactions of the Royal Society B: Biological Sciences, 363, 2841–2850. addition, if dispersal were responsible for these cyclical outbreaks, Anderson, E. C., & Thompson, E. A. (2002). A model-­based method for such frequent dispersal events would have long since obliterated identifying species hybrids using multilocus genetic data. Genetics, 160, the genetic distinction we discovered between eastern and west- 1217–1229. ern populations resulting in the predominance of the eastern genetic Babik, W., Branicki, W., Crnobrnja-Isailović, J., Cogălniceanu, D., Sas, I., Olgun, K., … Arntzen, J. W. (2005). Phylogeography of European newt lineage. In contrast, our simulation results suggest that the western species – discordance between mtDNA and morphology. Molecular population will become predominant after another ~2,000 years of Ecology, 14, 2475–2491. continued secondary contact, although populations of the eastern Babik, W., Pabijan, M., Arntzen, J. W., Cogălniceanu, D., Durka, W., & genetic lineage might likely still persist in the Caucasus region due to Radwan, J. (2009). Long-term­ survival of a urodele amphibian despite depleted major histocompatibility complex variation. Molecular Ecology, geographic isolation. 18, 769–781. Beck, J. B., Schmuths, H., & Schaal, B. A. (2007). Native range genetic vari- ation in Arabidopsis thaliana is strongly geographically structured and ACKNOWLEDGEMENTS reflects Pleistocene glacial dynamics. Molecular Ecology, 17, 902–915. The authors would like to thank the many individuals who aided in Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate -­ a practical and powerful approach to multiple testing. Journal of the Royal the collection of winter moths used in this study: Austria—G. Hoch Statistical Society Series B-­Methodological, 57, 289–300. and A. Schopf; Czech Republic and Slovakia—M. Turcáni; France—C. Buse, A., & Good, J. E. G. (1996). Synchronization of larval emergence Bonnet, F. Hérard, and A. Roques; Germany—M. Kenis, and F. Kruger; in winter moth (Operophtera brumata L) and budburst in peduncu- Georgia—G. Japoshvili; Italy—A. Battisti; Norway—J. Jepsen and T. late oak (Quercus robur L) under simulated climate change. Ecological Entomology, 21, 335–343. Scott; Poland—L. Sukovata; Serbia—M. Glavendekić; Spain—M. Canestrelli, D., Cimmaruta, R., Costantini, V., & Nascetti, G. (2006). Genetic Lombadero and A. Pepi; Sweden—H. Bylund; Switzerland—M. Kenis; diversity and phylogeography of the Apennine yellow-­bellied toad United Kingdom—H. Evans, C. Tilbury, A. Vanbergen, and A. Watt. Bombina pachypus, with implications for conservation. Molecular We would also like to thank D. Newman for laboratory assistance, Ecology, 15, 3741–3754. Chapuis, M.-P., & Estoup, A. (2007). Microsatellite null alleles and estima- Dr. N.J. Mills for providing laboratory access, and G. Boettner, tion of population differentiation. Molecular Biology and Evolution, 24, H. Broadley, M. Davis, R. Gwiazdowski, M. Labbé, J. Lombardo, T. 621–631. Murphy, and R. Crandall for their creative comments and suggestions Clark, P. U., Dyke, A. S., Shakun, J. D., Carlson, A. E., Clark, J., Wohlfarth, B., … throughout this study. We would also like to thank Dr. A. Beckerman McCabe, A. M. (2009). The last glacial maximum. Science, 325, 710–714. Cornuet, J.-M., Pudlo, P., Veyssier, J., Dehne-Garcia, A., Gautier, M., Leblois, and four anonymous reviewers whose comments greatly improved R., … Estoup, A. (2014). DIYABC v2.0: A software to make approximate this manuscript. Bayesian computation inferences about population history using sin- gle nucleotide polymorphism, DNA sequence and microsatellite data. Bioinformatics, 30, 1187–1189. DATA ACCESSIBILITY Cornuet, J.-M., Santos, F., Beaumont, M. A., Robert, C. P., Marin, J.-M., Balding, D. J., … Estoup, A. (2008). Inferring population history with DIY Microsatellite genotype scores for each individual are provided as a ABC: A user-­friendly approach to approximate Bayesian computation. tab-­delimited file in single line SetEuceture format. See Appendix S2. Bioinformatics, 24, 2713–2719. Crawford, N. G. (2010). smogd: Software for the measurement of genetic diversity. Molecular Ecology Resources, 10, 556–557. AUTHOR CONTRIBUTIONS Derks, M. F. L., Smit, S., Salis, L., Schijlen, E., Bossers, A., Mateman, C., … Megens, H.-J. (2015). The genome of winter moth (Operophtera bru- JSE coordinated the collection of samples and directed the research. mata) provides a genomic perspective on sexual dimorphism and phe- AC provided laboratory access and oversaw the analyses. JCA and nology. Genome Biology and Evolution, 7, 2321–2332. NPH collected the molecular data. JCA analyzed the dataset and over- Dubey, S., Zaitsev, M., Cosson, J.-F., Abdukadier, A., & Vogel, P. (2006). Pliocene and Pleistocene diversification and multiple refugia in a saw manuscript preparation. All authors contributed to the project de- Eurasian shrew (Crocidura suaveolens group). Molecular Phylogenetics sign and in preparing the manuscript. and Evolution, 38, 635–647. ANDERSEN et al | 3322

Dumolin-Lapègue, S., Demesure, B., Fineschi, S., LeCorre, V., & Petit, R. Hunter, M. D., Watt, A. D., & Docherty, M. (1991). Outbreaks of winter moth J. (1997). Phylogeographic structure of white oaks throughout the on Sitka Spruce in Scotland are not influenced by nutrient deficiencies European continent. Genetics, 146, 1475–1487. of trees, tree budburst, or pupal predation. Oecologia, 86, 62–69. Edger, P. P., Heidel-Fischer, H. M., Bekaert, M., Rota, J., Glöckner, G., Platts, Huson, D. H., & Bryant, D. (2006). Application of phylogenetic networks A. E., … Wheat, C. W. (2015). The butterfly plant arms-­race escalated in evolutionary studies. Molecular Biology and Evolution, 23, 254–267. by gene and genome duplications. Proceedings of the National Academy Jaarola, M., & Searle, J. B. (2002). Phylogeography of field voles (Microtus of Sciences of the United States of America, 112, 8362–8366. agrestis) in Europe inferred from mitochondrial DNA sequences. Elias, M., Faria, R., Gompert, Z., & Hendry, A. (2012). Factors influencing Molecular Ecology, 11, 2613–2621. progress toward ecological speciation. International Journal of Ecology, Jepsen, J. U., Vindstad, O. P. L., Barraquand, F., Ims, R. A., & Yoccoz, N. G. 2012, 1–7. (2016). Continental-­scale travelling waves in forest geometrids in Europe: Elkinton, J., Boettener, G., Liebhold, A., & Gwiazdowski, R. (2015). Biology, An evaluation of the evidence. Journal of Ecology, 85, 385–390. Spread, and Biological Control of Winter Moth in the Eastern United States. Kearse, M., Moir, R., Wilson, A., Stones-Havas, S., Cheung, M., Sturrock, S., Morgantown, WV: USDA Forest Service Publication FHTET-2014-07. … Drummond, A. (2012). Geneious Basic: An integrated and extend- Elkinton, J. S., Boettner, G. H., Sremac, M., Gwiazdowski, R., Hunkins, able desktop software platform for the organization and analysis of R. R., Callahan, J., … Campbell, N. K. (2010). Survey for winter moth sequence data. Bioinformatics, 28, 1647–1649. (Lepidoptera: Geometridae) in northeastern North America with Leggett, H. C., Jones, E. O., Burke, T., Hails, R. S., Sait, S. M., & Boots, M. pheromone-­baited traps and hybridization with the native Bruce span- (2011). Population genetic structure of the winter moth, Operophtera worm (Lepidoptera: Geometridae). Annals of the Entomological Society brumata Linnaeus, in the Orkney Isles suggests long-­distance dispersal. of America, 103, 135–145. Ecological Entomology, 36, 318–325. Elkinton, J. S., Lance, D., Boettner, G., Khrimian, A., & Leva, N. (2011). Liebhold, A., Koenig, W. D., & Bjornstad, O. N. (2004). Spatial synchrony Evaluation of pheromone-­baited traps for winter moth and Bruce span- in population dynamics. Annual Review of Ecology Evolution and worm (Lepidoptera: Geometridae). Journal of Economic Entomology, Systematics, 35, 467–490. 104, 494–500. Macphee, A., Newton, A., & McRae, K. B. (1988). Population studies on Elkinton, J. S., Liebhold, A., Boettner, G. H., & Sremac, M. (2014). Invasion the winter moth Operophtera-brumata (L) (Lepidoptera, Geometridae) spread of Operophtera brumata in northeastern United States and in apple orchards in Nova-­Scotia. Canadian Entomologist, 120, 73– ­hybridization with O-bruceata. Biological Invasions, 16, 2263–2272. 83. Falush, D., Stephens, M., & Pritchard, J. K. (2003). Inference of population Mannai, Y., Ezzine, O., Nouira, S., & Ben Jamaa, M. L. (2015). First report of the structure using multilocus genotype data: Linked loci and correlated winter moth Operophtera brumata on Quercus canariensis and Q. afares in allele frequencies. Genetics, 164, 1567–1587. North West of Tunisia. Tunisian Journal of Plant Protection, 10, 69–73. Feder, J. L., Berlocher, S. H., Roethele, J. B., Dambroski, H., Smith, J. J., Perry, Meirmans, P. G., & Van Tienderen, P. H. (2004). GENOTYPE and GENODIVE: W. L., … Aluja, M. (2003). Allopatric genetic origins for sympatric host-­ Two programs for the analysis of genetic diversity of asexual organ- plant shifts and race formation in Rhagoletis. Proceedings of the National isms. Molecular Ecology Notes, 4, 792–794. Academy of Sciences of the United States of America, 100, 10314–10319. Mezzasalma, M., Dall’Asta, A., Loy, A., Cheylan, M., Lymberakis, P., Zuffi, Ferguson, D. C. (1978). Pests not known to occur in the United States or of M. A. L., … Guarino, F. M. (2015). A sisters’ story: Comparative phy- limited distribution. Winter moth Operophtera brumata (L.) Lepidoptera: logeography and of Hierophis viridiflavus and H. gemonensis Geometridae. Cooperative Plant Pest Report, 3, 687–694. (Serpentes, Colubridae). Zoologica Scripta, 44, 495–508. Gwiazdowski, R. A., Elkinton, J. S., Dewaard, J. R., & Sremac, M. (2013). Peltonen, M., Liebhold, A. M., Bjornstad, O. N., & Williams, D. W. (2002). Phylogeographic diversity of the winter moths Operophtera brumata Spatial synchrony in forest insect outbreaks: Roles of regional stochas- and O. bruceata (Lepidoptera: Geometridae) in Europe and North ticity and dispersal. Ecology, 83, 3120–3129. America. Annals of the Entomological Society of America, 106, 143–151. Price, P. W., & Hunter, M. D. (2015). Population dynamics of an insect her- Habel, J. C., Lens, L., Rodder, D., & Schmitt, T. (2011). From Africa to Europe bivore over 32 years are driven by precipitation and host-­plant effects: and back: Refugia and range shifts cause high genetic differentiation Testing model predictions. Environmental Entomology, 44, 463–473. in the Marbled White butterfly Melanargia galathea. BMC Evolutionary Pritchard, J. K., Stephens, M., & Donnelly, P. (2000). Inference of population Biology, 11, 215. structure using multilocus genotype data. Genetics, 155, 945–959. Hampe, A., Arroyo, J., Jordano, P., & Petit, R. J. (2003). Rangewide phyloge- R Core Team (2015). R: A language and environment for statistical comput- ography of a bird-­dispersed Eurasian shrub: Contrasting Mediterranean ing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved and temperate glacial refugia. Molecular Ecology, 12, 3415–3426. from http://www.R-project.org/ Harrison, R. G., & Larson, E. L. (2014). Hybridization, introgression, and the Rambaut, A., & Drummond, A. (2007). Tracer v1.4, Retrieved from http:// nature of species boundaries. Journal of Heredity, 105, 795–809. beast.bio.ed.ac.uk/Tracer Hassell, M. P. (1968). Behavioural response of a tachinid fly (Cyzenis albi- Raymond, M., & Rousset, F. (1995). GENEPOP (version-­1.2) -­ Population-­ cans (Fall)) to its host winter moth (Operophtera brumata (L)). Journal of genetics software for exact tests and ecumenicism. Journal of Heredity, Animal Ecology, 37, 627–639. 86, 248–249. Havill, N. P., Elkinton, J. S., Andersen, J. C., Hagen, S. B., Broadley, H. J., von Reumont, B. M., Struwe, J.-F., Schwarzer, J., & Misof, B. (2012). Boettner, G. J., & Caccone, A. (2017). Asymmetric hybridization be- Phylogeography of the burnet moth Zygaena transalpina complex: tween non-­native winter moth, Operophtera brumata (Lepidoptera: Molecular and morphometric differentiation suggests glacial refugia in Geometridae), and native Bruce spanworm, O. bruceata, in the north- Southern France, Western France and micro-­refugia within the Alps. eastern United States, assessed with novel microsatellites and SNPs. Journal of Zoological Systematics and Evolutionary Research, 50, 38–50. Bulletin of Entomological Research, 107, 241–250. Riolo, M. A., Rohani, P., & Hunter, M. D. (2015). Local variation in plant qual- Hewitt, G. M. (1996). Some genetic consequences of ice ages, and their ity influences large-­scale population dynamics. Oikos, 124, 1160–1170. role in divergence and speciation. Biological Journal of the Linnean Roelofs, W. L., Hill, A. S., Linn, C. E., Meinwald, J., Jain, S. C., Herbert, H. J., Society, 58, 247–276. & Smith, R. F. (1982). Sex-­pheromone of the winter moth, a geometrid Hewitt, G. (2000). The genetic legacy of the Quaternary ice ages. Nature, with unusually low-­temperature pre-­copulatory responses. Science, 405, 907–913. 217, 657–659. Hunter, A. F., & Elkinton, J. S. (2000). Effects of synchrony with host plant on Rousselet, J., Zhao, R., Argal, D., Simonato, M., Battisti, A., Roques, A., populations of a spring-­feeding Lepidopteran. Ecology, 81, 1248–1261. & Kerdelhue, C. (2010). The role of topography in structuring the ANDERSEN et al | 3323

demographic history of the pine processionary moth, Thaumetopoea Van Dongen, S., Backeljau, T., Matthysen, E., & Dhondt, A. A. (1997). pityocampa (Lepidoptera: Notodontidae). Journal of Biogeography, 37, Synchronization of hatching date with budburst of individual host trees 1478–1490. (Quercus robur) in the winter moth (Operophtera brumata) and its fitness Rousset, F. (2008). GENEPOP ‘ 007: A complete re-­implementation of consequences. Journal of Animal Ecology, 66, 113–121. the GENEPOP software for Windows and Linux. Molecular Ecology Van Dongen, S., Backeljau, T., Matthysen, E., & Dhondt, A. A. (1998). Genetic pop- Resources, 8, 103–106. ulation structure of the winter moth (Operophtera brumata L.) (Lepidoptera, Schmitt, T. (2007). Molecular biogeography of Europe: Pleistocene cycles Geometridae) in a fragmented landscape. Heredity, 80, 92–100. and postglacial trends. Frontiers in Zoology, 4, 1–13. Van Dongen, S., Matthysen, E., & Dhondt, A. A. (1996). Restricted male Schmitt, T., & Müller, P. (2007). Limited hybridization along a large con- winter moth (Operophtera brumata L) dispersal among host trees. Acta tact zone between two genetic lineages of the butterfly Erebia me- Oecologica-International­ Journal of Ecology, 17, 319–329. dusa (Satyrinae, Lepidoptera) in Central Europe. Journal of Zoological Varley, G. C., & Gradwell, G. R. (1960). Key factors in population studies. Systematics and Evolutionary Research, 45, 39–46. Journal of Animal Ecology, 29, 399–401. Schmitt, T., Rober, S., & Seitz, A. (2005). Is the last glaciation the only rele- Varley, G. C., & Gradwell, G. R. (1968). Population models for the winter vant event for the present genetic population structure of the meadow moth. Insect abundance, 46, 132–142. brown butterfly Maniola jurtina (Lepidoptera: Nymphalidae)? Biological Verhoeven, K. J. F., Macel, M., Wolfe, L. M., & Biere, A. (2011). Population Journal of the Linnean Society, 85, 419–431. admixture, biological invasions and the balance between local adap- Schmitt, T., & Seitz, A. (2001a). Allozyme variation in Polyommatus coridon tation and inbreeding depression. Proceedings of the Royal Society B: (Lepidoptera: Lycaenidae): Identification of ice-­age refugia and reconstruc- Biological Sciences, 278, 2–8. tion of post-­glacial expansion. Journal of Biogeography, 28, 1129–1136. Vinstad, O. P. L., Schott, T., Hagen, S. B., Jepsen, J. U., Kapari, L., & Ims, Schmitt, T., & Seitz, A. (2001b). Intraspecific allozymatic differentiation re- R. A. (2013). How rapidly do invasive birch forest geometrids recruit veals the glacial refugia and the postglacial expansions of European larval parasitoids? Insights from comparison with a sympatric native Erebia medusa (Lepidoptera: Nymphalidae). Biological Journal of the geometrid. Biological Invasions, 15, 1573–1589. Linnaean Society, 74, 429–458. Vitales, D., Garcia-Fernandez, A., Garnatje, T., Pellicer, J., & Valles, J. (2016). Schwenk, K., Brede, N., & Streit, B. (2008). Introduction. Extent, processes and Phylogeographic insights of the lowland species Cheirolophus semper- evolutionary impact of interspecific hybridization in . Philosophical virens in the southwestern Iberian Peninsula. Journal of Systematics and Transactions of the Royal Society B: Biological Sciences, 363, 2805–2811. Evolution, 54, 65–74. Seehausen, O., Takimoto, G., Roy, D., & Jokela, J. (2007). Speciation rever- Watt, A. D., & Mcfarlane, A. M. (1991). Winter moth on Sitka spruce -­ syn- sal and biodiversity dynamics with hybridization in changing environ- chrony of egg hatch and budburst, and its effect on larval survival. ments. Molecular Ecology, 17, 30–44. Ecological Entomology, 16, 387–390. Sinama, M., Dubut, V., Costedoat, C., Gilles, A., Junker, M., Malausa, Wielstra, B., Crnobrnja-Isailović, J., Litvinchuk, S. N., Reijnen, B. T., T., … Meglécz, E. (2011). Challenges of microsatellite development Skidmore, A. K., Sotiropoulos, K., … Arntzen, J. W. (2013). Tracing glacial in Lepidoptera: Euphydrays aurinia (Nymphalidae) as a case study. refugia of Triturus newts based on mitochondrial DNA phylogeography European Jouranal of Entomology, 108, 261–266. and species distribution modeling. Frontiers in Zoology, 10, 13. Stoakley, J. T. (1985). Outbreaks of winter moth, Operophtera brumata L. Wilson, G. A., & Rannala, B. (2003). Bayesian inference of recent migration (Lep., Geometridae) in young plantations of Sitka spruce in Scotland: rates using multilocus genotypes. Genetics, 163, 1177–1191. Insecticidal control and population assessment using sex attractant Wint, W. (1983). The role of alternative host-­plant species in the life of a pheromone. Journal of Applied Entomology, 99, 153–160. polyphagous moth, Operophtera-brumata (Lepidoptera, Geometridae). Svenning, J.-C., Eiserhardt, W. L., Normand, S., Ordonez, A., & Sandel, Journal of Animal Ecology, 52, 439–450. B. (2015). The influence of paleoclimate on present-­day patterns in Wood, S. N. (2011). Fast stable restricted maximum likelihood and marginal biodiversity and ecosystems. Annual Review of Ecology, Evolution, and likelihood estimations of semiparametric generalized linear models. Systematics, 46, 551–572. Journal of the Royal Statistical Society B, 73, 3–36. Taberlet, P., Fumagalli, L., Wust-Saucy, A. G., & Cosson, J. F. (1998). Wu, Y., Molongoski, J. J., Winograd, D. F., Bogdanowicz, S. M., Louyakis, A. S., Comparative phylogeography and postglacial colonization routes in Lance, D. R., Mastro, V. C., & Harrison, R. G. (2015). Genetic structure, Europe. Molecular Ecology, 7, 453–464. admixture and invasion success in a Holarctic defoliator, the gypsy moth Tenow, O. (2016). Continental-­scale travelling waves in forest geometrids (Lymantria dispar, Lepidoptera: Erebidae). Molecular Ecology, 24, 1275–1291. in Europe: An evaluation of the evidence Response. Journal of Animal You, M., Yue, Z., He, W., Yang, X., Yang, G., Xie, M., … Wang, J. (2013). A Ecology, 85, 391–395. heterozygous moth genome provides insights into herbivory and de- Tenow, O., Nilssen, A. C., Bylund, H., Pettersson, R., Battisti, A., Bohn, U., toxification. Nature Genetics, 45, 220–225. … Utkina, I. (2013). Geometrid outbreak waves travel across Europe. Journal of Animal Ecology, 82, 84–95. Tikkanen, O. P., & Lyytikainen-Saarenmaa, P. (2002). Adaptation of a gen- SUPPORTING INFORMATION eralist moth, Operophtera brumata, to variable budburst phenology of host plants. Entomologia Experimentalis Et Applicata, 103, 123–133. Additional Supporting Information may be found online in the sup- Tikkanen, O. P., Woodcock, B., Watt, A., & Lock, K. (2006). Are polyphagous porting information tab for this article. geometrid moths with flightless females adapted to budburst phenol- ogy of local host species? Oikos, 112, 83–90. Tison, J.-L., Edmark, V. N., Sandoval-Castellanos, E., Van Dyck, H., Tammaru, How to cite this article: Andersen JC, Havill NP, Caccone A, T., Valimaki, P., … Gotthard, K. (2014). Signature of post-­glacial expan- Elkinton JS. Postglacial recolonization shaped the genetic diversity sion and genetic structure at the northern range limit of the speckled wood butterfly. Biological Journal of the Linnean Society, 113, 136–148. of the winter moth (Operophtera brumata) in Europe. Ecol Evol. Troubridge, J. T., & Fitzpatrick, S. M. (1993). A revision of the North-­ 2017;7:3312–3323. https://doi.org/10.1002/ece3.2860 American Operophtera (Lepidoptera, Geometridae). Canadian Entomologist, 125, 379–397.